Detection of Rainfall Probability Distribution from Atmospheric Parameters Using the Neural Network Modelling Algorithm
Okoro, N. O. and Ogbonna R. C
Keywords: Atmospheric parameters, modelling algorithm, neural network, rainfall distribution probability.
Abstract
The Earth’s surface receives different amounts of rainfall during one season in different places and that too during another season throughout the year. It is a priority in Nigeria to detect and understand spatiotemporal rainfall distribution probabilities to anticipate suitable strategies for environmental hazard monitoring and control, agricultural productivity and water management control. In Nigeria, obtaining data and information regarding rainfall distribution probability has been strongly dependent on the use of rainfall forecasting instruments by Nigeria Meteorological Agency (NiMET). This has made information dissemination about rainfall so difficult to reach out to many people especially the ruler dwellers. Therefore, the objective of this study is to use neural network modelling algorithms and atmospheric parameters (temperature, pressure, relative humidity, wind speed and total cloud cover) to detect rainfall probability distributions (that is, when to and when not to expect rainfall) at any given location. A programme was designed using a neural network modelling algorithm to permit the input of atmospheric parameter(s) for the detection of rainfall probability distributions within the area of study for different periods of time. The results obtained showed that rainfall distribution tendencies on different days in the year of study corresponded with the observed rainfall data distribution for the same year of study within the same location. This clearly indicated that the rainfall probability distribution can possibly be detected from the temperature, pressure, relative humidity, wind speed and total cloud cover using a neural network modelling algorithm. This may have implications for agricultural productivity and sustainability when monitoring rainfall tendencies during the actual planting period. It may also be useful in monitoring rainfall activities in any given location to reduce hazards.